1. Restoration actions can halt biodiversity loss and rescue its services. However, in order to be effective, priority areas for restoration should be chosen based on objective large-scale restoration planning. Here, a multi-criteria graph theory (GT) framework is proposed to indicate priority areas for active restoration, based on landscape resilience, landscape connectivity, and biodiversity conservation value, focusing on threatened endemic plant species.

2. We applied this GT framework to 10,406 catchment basins of the Brazilian Caatinga, the largest seasonally dry tropical forest of the New World. Vegetation cover and within-catchment connectivity were used to identify catchments of intermediate landscape resilience, which in principle offer more effective opportunities for restoration. Then, such catchments were independently classified into (i) three classes according their value for between-catchment connectivity and (ii) three classes of biodiversity conservation value, based on richness of threatened, endemic plant species. By the integration of landscape resilience, landscape connectivity and biodiversity conservation values, three priority classes for restoration were generated.

3. The multi-criteria framework generated several restoration priority cut-offs. Prioritization based on landscape resilience selected 36% of the Caatinga catchments as high priority for restoration. By independently adding landscape connectivity and biodiversity conservation value, only 12% and 3% of the catchments, respectively, were considered high priority. By combining all three criteria, 9% of the catchments were selected as high priority and less than 1% as top priority for restoration.

4. Synthesis and applications: The multicriteria GT framework for restoration prioritization, which maximizes the effectiveness of restoration actions, landscape connectivity for climate change adaptation and conservation of threatened species, can be applied worldwide under different budged limitations and spatial scales, being useful for private, state, and federal initiatives.

The 2009 Caatinga vegetation cover map used as the basis of this work. It was he national official data, in that occasion, and was made available as georeferenced digital files (shapefiles) in 2014 by the Satellite Deforestation Monitoring Project (PMDBBS) of the Brazilian Ministry of Environment (MMA). Another database of this work was the water catchment basin, delimited by the authors using ArcHydro 2.0 module for ArcGis® 10.1 based on SRTM dataset (version 4 - http://srtm.csi.cgiar.org/). To improve the solution, the original SRTM raster file was reconditioned using as reference the digitized drainage vector layer from Instituto Brasileiro de Geografia e Estatística (IBGE) at 1:250,000 scale. Sub basin grids with threshold of 7,500 ha were created for the entire Caatinga region using the flow direction grid, segmented stream grids developed at this scale, and the Hydro Tools “catchment grid delineation” tool. These sub-basin grids were then converted into separate polygon shapefiles using the Hydro Tools “catchment polygon processing” tool. The detailed description of the data source and the bases derivation are described in Antongiovanni et al. accepted to be published in Journal of Applied Ecology. The detailed description of the variables used in this dataset and how they were generated are also avaiable in the metadata below.

The detailed description of the variables used in this dataset and how they were generated are in the table below.